With the development of communications technologies, ALL Internet Protocol (IP) network has gradually become reality in network construction of operators. Every operator has basically stopped newly building a traditional voice switching network and existing service also gradually starts to be transformed into the direction of an IP carrier. At present, an IP network in the related art is a “best-effort” service model, and neither the quality of service nor the reliability can achieve levels of a traditional telecommunications network. The telecommunications network, which is more sensitive to the reliability of a link, generally requires that switching of a host link and a backup link can be implemented within 50 milliseconds (ms) when a link fails. That is, a service is switched from the host link, which fails, to the backup link within 50 ms. Various link detection protocols, including Bidirectional Forwarding Detection (BFD), Operations, Administration and Maintenance (OAM) of Multiprotocol Label Switching-Transport Profile (MPLS-TP), and so on and various link switching mechanisms have been put forward in the industry in order to solve the above technical problem.
Generally, each host link corresponds to a backup link in the telecommunications network, and the two links form a host-backup relation, that is a protection group. When the host link fails, a service is switched to the backup link. A link detection protocol session, such as BFD, and MPLS-TP OAM, is configured on the host link. Such detection protocols, which are highly sensitive, are able to detect a failure of the host link in time. The detection protocols, which are required to be highly sensitive, need to be supported by one or more hardware chips and are generally implemented on one or more line cards. A warning is generated when the link detection protocol session detects a failure of the host link. Information of the warning may be notified to a Central Processing Unit (CPU) on a line card directly so as to perform link switching. Of course, the information of the warning may be also firstly notified to the CPU on the line card, subsequently notified by the CPU on the line card to a CPU on a main control board, and finally notified by the CPU on the main control board to the CPU on the line card to perform the switching. Apparently, the former method is fully adequate to complete link switching when there is only one line card in a system, but is only able to complete link switching where a session locates when there is a plurality of line cards in the system, thus failing to complete link switching on other line cards.
Generally, a high-end device in a network applies a fully distributed architecture, mainly consisting of a main control board, one or more line cards and a switching network. The main control board is configured to take charge of a system control plane. Each line card is configured to take charge of a control plane and a data forwarding plane of a line card itself. A data forwarding service between the one or more line cards is performed by the switching network. The main controller board communicates with the one or more line cards via one or more messages. In a distributed system, there is an entry of a forwarding table of a data forwarding plane on each line card for a multipoint-to-multipoint service. Through the former method, although switching is performed by a forwarding link on a line card where the link detection session locates, services on other line cards are still operating on the host link where a failure occurs. Therefore, only the latter switching method can be used in the distributed system. Link switching is performed directly by the line card where the link detection session locates while the main controller board is notified. The main controller board then synchronizes switching information to other line cards, and the other line cards perform link switching after receiving a switching message. However, such a method has a obvious disadvantage, that is, the efficiency of switching depends on the efficiency of synchronizing a message between boards while the efficiency of synchronizing the message between boards is uncontrollable. Besides, a message may be also lost if the message is not processed in time when there are a large number of messages between boards.